bug#23559: 25.0.93; Wrong type argument on C-u C-x = with Czech input method

bug#23559: 25.0.93; Wrong type argument on C-u C-x = with Czech input method

[Milan Zamazal]

When I press `C-u C-x =' with a Czech input method enabled, I get the
following error:

  quail-gen-decode-map: Wrong type argument: characterp, kp-add

How to reproduce the bug:

- Start Emacs, e.g.: emacs -Q
- Select Czech input method: C-x RET C-\ czech RET
- Make sure you are on some character in the buffer, e.g. go to the
  beginning of *scratch* buffer with the semicolon character.
- Run what-cursor-position with a prefix argument: C-u C-x =
- You should receive the error above.

The bug is also present in Emacs 24.5.

bug#23559: 25.0.93; Wrong type argument on C-u C-x = with Czech input method

[Eli Zaretskii]

> From: Milan Zamazal <pdm@zamazal.org>
> Date: Thu, 12 May 2016 21:35:55 +0200
> 
> When I press `C-u C-x =' with a Czech input method enabled, I get the
> following error:
> 
>   quail-gen-decode-map: Wrong type argument: characterp, kp-add

Do you (or someone else) understand why Czech and Slovak input methods
include kp-* keys in the input method definitions?  Like this:

(quail-define-rules
 ("1" ?+)
 ...
 ("=}" ?})
 ([kp-1] ?1)
 ([kp-2] ?2)
 ([kp-3] ?3)
 ([kp-4] ?4)
 ([kp-5] ?5)
 ([kp-6] ?6)
 ([kp-7] ?7)
 ([kp-8] ?8)
 ([kp-9] ?9)
 ([kp-0] ?0)
 ([kp-add] ?+))

What does this try to accomplish, and why?  No other input method does
that, and the bindings this tries to produce are already available by
default, so what is their purpose?  Those bindings are the cause of
this problem, AFAICT.

Thanks.

bug#23559: 25.0.93; Wrong type argument on C-u C-x = with Czech input method

[Milan Zamazal]

>>>>> "EZ" == Eli Zaretskii <eliz@gnu.org> writes:

    EZ> Do you (or someone else) understand why Czech and Slovak input
    EZ> methods include kp-* keys in the input method definitions?

The reason is that those input methods replace digit keys at the main
part of the keyboard with Czech and Slovak characters.  But the keys at
the numeric keypad should of course still produce digits and not Czech
and Slovak characters.

IIRC at the time those input methods were introduced, quail definitions
like

  ("1" ?+)

caused both the `1' and kp-1 keys to produce `+' and adding

  ([kp-1] ?1)

was necessary to keep kp-1 producing `1' and not `+'.  If this is no
longer necessary then those bindings can be simply removed.

bug#23559: 25.0.93; Wrong type argument on C-u C-x = with Czech input method

[Eli Zaretskii]

> From: Milan Zamazal <pdm@zamazal.org>
> Cc: 23559@debbugs.gnu.org
> Date: Thu, 19 May 2016 20:50:33 +0200
> 
> >>>>> "EZ" == Eli Zaretskii <eliz@gnu.org> writes:
> 
>     EZ> Do you (or someone else) understand why Czech and Slovak input
>     EZ> methods include kp-* keys in the input method definitions?
> 
> The reason is that those input methods replace digit keys at the main
> part of the keyboard with Czech and Slovak characters.  But the keys at
> the numeric keypad should of course still produce digits and not Czech
> and Slovak characters.
> 
> IIRC at the time those input methods were introduced, quail definitions
> like
> 
>   ("1" ?+)
> 
> caused both the `1' and kp-1 keys to produce `+' and adding
> 
>   ([kp-1] ?1)
> 
> was necessary to keep kp-1 producing `1' and not `+'.

I see, thanks.

> If this is no longer necessary then those bindings can be simply
> removed.

They don't seem to be necessary, so I removed them, and I'm marking
the bug done.

Thanks.